<robclark>
alanc: disregard my last statement... even if they remain embargoed nonsense forever distros still see them.. but maybe we need some clear doc (either in drm kernel docs or mesa) about security boundaries and how UMD isn't one (see https://docs.vulkan.org/guide/latest/validation_overview.html#_undefined_behavior ... but also that is just making explicit for vk what is implicit for legacy APIs, ie gl/es, etc)... that would at
<robclark>
least give us something to copy/paste link to to argue against nonsense CVE's rather than have to spend time...
ajhalaney[m] has quit []
Andy[m] has quit []
Celmor[m] has quit []
hch12907 has quit []
zzxyb[m] has quit []
neobrain[m] has quit []
heat_ has quit [Ping timeout: 480 seconds]
Ella[m] has quit []
krushia has quit [Quit: Konversation terminated!]
bylaws1 has quit []
sythemeta847[m] has quit []
Leopold has quit [Remote host closed the connection]
general_j[m] has quit []
underpantsgnome[m] has quit []
Haaninjo has joined #dri-devel
zackr has joined #dri-devel
glennk has quit [Ping timeout: 480 seconds]
nielsdg has quit []
yyds has joined #dri-devel
Haaninjo has quit [Ping timeout: 480 seconds]
kzd has quit [Quit: kzd]
Soroush has quit []
MarkCollins[m] has quit []
pac85[m] has quit []
lemonzest has quit [Quit: WeeChat 4.2.1]
gallo[m] has quit []
Labnan[m] has quit []
mairacanal[m] has quit []
lemonzest has joined #dri-devel
Leopold has joined #dri-devel
camus1 has quit [Remote host closed the connection]
Leopold has quit [Remote host closed the connection]
camus has joined #dri-devel
eballetbo has quit []
matrix638[m] has quit []
dafna33[m] has quit []
Leopold_ has joined #dri-devel
kts has joined #dri-devel
Leopold_ has quit [Remote host closed the connection]
<DavidHeidelberg>
karolherbst: around? I'm trying to use cl.clCreateImage2D in tinygrad which supposedly should work, but on both freedreno and iris it fails with -40 (cannot allocate). I looked at gallium properties and rust code and it looks like it should work, but somehow it doesn't
simon-perretta-img has quit [Ping timeout: 480 seconds]
<karolherbst>
DavidHeidelberg: -40 as the CL error code?
<karolherbst>
-40 is CL_INVALID_IMAGE_SIZE
simon-perretta-img has joined #dri-devel
<karolherbst>
usually means that the image is bigger than supported by the hardware
<karolherbst>
is checked inside `validate_image_desc`
<DavidHeidelberg>
yeah, but both iris and fd has max caps (at least it seems to me (16384)
<DavidHeidelberg>
1st alloc: 12565x6
<DavidHeidelberg>
2nd: 50260x192
<DavidHeidelberg>
3rd: 50260x192
<karolherbst>
DavidHeidelberg: all fail?
<karolherbst>
ehh
<karolherbst>
2nd and 3rd should fail
<DavidHeidelberg>
I assume just the 3rd fails
<karolherbst>
yeah, that's not supported by the drivers
<DavidHeidelberg>
the status is read afterwards, so...
simon-perretta-img has quit [Ping timeout: 480 seconds]
<karolherbst>
yeah.. looks like rusticl behaves correctly here then, need to split the image on the application side
<karolherbst>
or just use a raw buffer
crabbedhaloablut has quit []
krei-se has joined #dri-devel
crabbedhaloablut has joined #dri-devel
jernej_ has joined #dri-devel
<tnt>
or use 12565 x 768 . I guess the GPU uses tiling for better caching when using images vs buffer and you might want to keep that behavior.
Leopold_ has joined #dri-devel
jernej has quit [Read error: Connection reset by peer]
<DavidHeidelberg>
got suggestion on Adreno there is more efficient to use 2dimage because of cache
<DavidHeidelberg>
*faster cache for it
<karolherbst>
yeah... makes sense
<karolherbst>
then you kinda have to split the image if you need images as huge as those
sebastiencs9 has quit []
<karolherbst>
could also use a 2darray thing and slice it :D
<karolherbst>
gallium uses 16 bit in a couple of places, so I'm not quite sure we won't run into weirdo issues as some of the values are signed
<karolherbst>
so we might get to 32x32 easily, but 64x64 might be difficult
ella-0[m] has joined #dri-devel
ohadsharabi[m] has joined #dri-devel
kallisti5[m] has joined #dri-devel
tomeu has joined #dri-devel
treeq[m] has joined #dri-devel
ofirbitt[m] has joined #dri-devel
heftig has joined #dri-devel
tak2hu[m] has joined #dri-devel
tuxayo has joined #dri-devel
dantob has joined #dri-devel
Tooniis[m] has joined #dri-devel
Coelacanthus[m]1 has joined #dri-devel
LinuxHackerman has joined #dri-devel
enunes[m] has joined #dri-devel
tleydxdy has joined #dri-devel
<DavidHeidelberg>
hmm, well, if we get closer to the usefulness of proprietary drivers, why not. I'll just do quick test on freedreno if it'll pass the texture..
enick_936 has joined #dri-devel
m00nlit[m] has joined #dri-devel
aradhya7[m] has joined #dri-devel
kelbaz[m] has joined #dri-devel
nyorain[m] has joined #dri-devel
zamundaaa[m] has joined #dri-devel
naheemsays[m] has joined #dri-devel
Anson[m] has joined #dri-devel
K0bin[m] has joined #dri-devel
michael5050[m] has joined #dri-devel
koki23[m] has joined #dri-devel
moben[m] has joined #dri-devel
nick1343[m] has joined #dri-devel
kunal_10185[m] has joined #dri-devel
chema has joined #dri-devel
AlexisHernndezGuzmn[m] has joined #dri-devel
Targetball[m] has joined #dri-devel
YaLTeR[m] has joined #dri-devel
exp80[m] has joined #dri-devel
Eighth_Doctor has joined #dri-devel
kos_tom has joined #dri-devel
<karolherbst>
yeah.. the question is if we need new caps or not
<karolherbst>
but I susepct we will
onox[m] has joined #dri-devel
QiuWenbo[m] has joined #dri-devel
Wallbraker has joined #dri-devel
sigmoidfunc[m] has joined #dri-devel
<DavidHeidelberg>
the regular textures keep the same limit.. so, me too
znullptr[m] has joined #dri-devel
cwfitzgerald[m] has joined #dri-devel
kusma has joined #dri-devel
Newbyte has joined #dri-devel
<jenatali>
IIRC you run into issues sampling at that resolution too, not enough bits for subpixel precision or something like that
siddh has joined #dri-devel
aura[m] has joined #dri-devel
<karolherbst>
some drivers just scale internally to support sampled images
<karolherbst>
(well.. the ISA might also have restrictions there)
EricCurtin[m] has joined #dri-devel
PiGLDN[m] has joined #dri-devel
robertmader[m] has joined #dri-devel
KunalAgarwal[m][m] has joined #dri-devel
FL4SHK[m] has joined #dri-devel
<jenatali>
I know the D3D spec says texture coords go to 16.8 fixed point, so if the size is greater than 16 bit that has to change and you lose subtexel precision
<tnt>
yeah, although I'm realizing my kernel isn't all that recent so I should update first before dragging anyone into it :)
djbw has joined #dri-devel
<DavidHeidelberg>
karolherbst: btw. 65535x65535 passed on freedreno, sadly the code fails later (unrelated to the problem, but not fixed yet). At least the creation worked.
<karolherbst>
DavidHeidelberg: the issue is that pipe_box is all signed :)
<karolherbst>
and I'm sure people will complain changing that to unsigned or bumping to 32 bit because overhead on the GL side
<karolherbst>
so I'd rather make 32x32 work for now unless there are _strong_ reasons to bump it and go through all the work of making everybody happy here
<karolherbst>
there are also 32 bit limits on buffer sizes which is a more urgent issue and will probably also require some of that
<DavidHeidelberg>
right, the pipe_box making it clean it won't work :D
<tnt>
Does glsl has some function to check if a condition is true in any thread of a workgroup ? Something like anyInvocationARB but that works for the whole workgroup ?
<DavidHeidelberg>
karolherbst: but.. but... "x and width are used by buffers, so they need the full 32-bit range."
<karolherbst>
mhhhhh..
<karolherbst>
I wonder if we could expose 64x32....
<DavidHeidelberg>
yup :D
<DavidHeidelberg>
imho this would cover a lot, because still full 64x64k takes lot of VRAM even these days
<karolherbst>
yeah..
<karolherbst>
I guess freedreno has to report 16x16 for gl stilll, because MSAA and mipmaps?
<DavidHeidelberg>
sure, I think that apply for everyone in general
<DavidHeidelberg>
btw. my hack maybe will work, because it's "50260x192" texture
<DavidHeidelberg>
we could add pipe cap which will be used for CL and then for height just rusticl would MIN(32768,val)
<DavidHeidelberg>
would report
<karolherbst>
why does it need to be 50260x192 anyway?
<DavidHeidelberg>
I have totally no idea, I tried to read tinygrad code, but I would have to spend much more time to understand it
<karolherbst>
yeah.. it's not the best project...
<karolherbst>
and not even fast
<DavidHeidelberg>
imho it's good that it works relatively good with Mesa, on other hand, the hunt after getting low number of lines.. leads to very condensed code
<karolherbst>
yeah.. it's silly
<karolherbst>
I should fix those openvino bugs, but they are bugs in openvino.. :D
<karolherbst>
DavidHeidelberg: anything clblast based should also work btw
<karolherbst>
and that's actually fast
<karolherbst>
llama.cpp and whisper.cpp should just work (tm). I've tried the latter already. So if you can get the same thing done with something clblast based, I'd just use that
<DavidHeidelberg>
hmm, I'll have to find time to do more exploration :) I stayed with tinygrad since airlied published the blog